Type I interferon supports primary CD8⁺ T-cell responses to peptide-pulsed dendritic cells in the absence of CD4⁺ T-cell help

Abstract

CD8⁺ T-cell responses to non-pathogen, cell-associated antigens such as minor alloantigens or peptide-pulsed dendritic cells (DC) are usually strongly dependent on help from CD4⁺ T cells. However, some studies have described help-independent primary CD8⁺ T-cell responses to cell-associated antigens, using immunization strategies likely to trigger natural killer (NK) cell activation and inflammatory cytokine production. We asked whether NK cell activation by MHC I-deficient cells, or administration of inflammatory cytokines, could support CD4⁺ T-cell help-independent primary responses to peptide-pulsed DC. Injection of MHC I-deficient cells cross-primed CD8⁺ T-cell responses to the protein antigen ovalbumin (OVA) and the male antigen HY, but did not stimulate CD8⁺ T-cell responses in CD4-depleted mice; hence NK cell stimulation by MHC I-deficient cells did not replace CD4⁺ T-cell help in our experiments. Dendritic cells cultured with tumour necrosis factor-α (TNF-α) or type I interferon-α (IFN-α) also failed to prime CD8⁺ T-cell responses in the absence of help. Injection of TNF-α increased lymph node cellularity, but did not generate help-independent CD8⁺ T-cell responses. In contrast, CD4-depleted mice injected with IFN-α made substantial primary CD8⁺ T-cell responses to peptide-pulsed DC. Mice deficient for the type I IFN receptor (IFNR1) made CD8⁺ T-cell responses to IFNR1-deficient, peptide-pulsed DC; hence IFN-α does not appear to be a downstream mediator of CD4⁺ T-cell help. We suggest that primary CD8⁺ T-cell responses will become help-independent whenever endogenous IFN-α secretion is stimulated by tissue damage, infection, or autoimmune disease.

Introduction

It is now widely accepted that CD4⁺ T-cell help is essential for the development of protective CD8⁺ T-cell immunity to many viral and bacterial pathogens, as well as cell-associated, non-inflammatory antigens.¹ Although the mechanisms are not yet fully understood, help can be important for many aspects of CD8⁺ T-cell stimulation, including generation of the initial response,^2–4 programming and maintaining the memory cell population,^5–9 and generation of protective memory/secondary responses.^10–16

Primary CD8⁺ T-cell responses are highly variable in their need for help.¹ Viral and bacterial pathogens usually stimulate strong, help-independent primary responses that can be measured directly ex vivo by classical cytotoxicity assays. In contrast, responses to cell-associated, minimally inflammatory antigens are usually less potent, and cannot be detected by traditional cytotoxicity assays without antigen restimulation for several days in vitro. Because these secondary cytotoxic responses are generally strongly dependent on stimulation of cognate help during the primary response, it was assumed for many years that the primary CD8⁺ T-cell responses were also help-dependent.

The development of more sensitive techniques to measure T-cell responses, such as tetramer staining, intracellular cytokine staining, and cytokine ELISPOT assays, revealed unexpected heterogeneity in the help-dependence of primary CD8 T-cell responses to cell-associated antigens. At one end of the spectrum, naive CD8⁺ T cells cross-primed by injection of antigen-expressing MHC I-deficient cells proliferated and differentiated into effector cells with or without help, even though unhelped secondary responses were severely impaired; this suggested that help was important for the generation of memory, but was not required for primary responses.⁷ In contrast, primary CD8⁺ T-cell responses stimulated directly by peptide-pulsed dendritic cells (DC), or cross-primed by antigen-loaded splenocytes, were strongly dependent on CD4⁺ T-cell help.^2–4 Comparing the different experimental models, we suggested that unhelped primary CD8⁺ T-cell responses might be driven either by help from endogenous natural killer (NK) cells stimulated by injection of MHC I-deficient cells,^17–19 or by endogenous inflammation triggered by the injection of large numbers of irradiated cells.

In this study, we investigated whether NK cell activation, or exposure to inflammatory cytokines, could drive help-independent CD8⁺ T-cell responses to peptide-pulsed DC. We found that NK cell activation by MHC I-deficient cells did not support primary CD8⁺ T-cell responses in the absence of CD4⁺ T-cell help, in agreement with a recent report.²⁰ In contrast, injection of interferon-α (IFN-α) – but not of another inflammatory cytokine, tumour necrosis factor-α (TNF-α) – supported CD8⁺ T-cell responses in CD4-depleted mice. The demonstration that type 1 interferon receptor (IFNR1) -deficient mice were able to make primary CD8⁺ T-cell responses when immunized with IFNR1-deficient, peptide-pulsed DC suggests that IFN-α does not replace CD4⁺ T-cell help, but instead supports primary CD8⁺ T-cell responses through an alternative mechanism.

Materials and methods

Mice

C57BL/6 (B6) and Tap-deficient²¹ mice were purchased from Taconic Farms (Germantown, NY) or the Jackson Laboratory (Bar Harbor, ME). The IFNRI-deficient mice²² were a generous gift from Dr Ray Welsh, Department of Pathology, University of Massachusetts School of Medicine. Mice were maintained in the specific pathogen-free vivarium at the University of Rochester Medical Center. All experiments were performed with the approval of the University of Rochester Institutional Animal Care and Use Committee.

Media

Cells were cultured in Iscove's modified Dulbecco's medium with 2 mm l-glutamine, 5 × 10⁻⁵ m 2-mercaptoethanol, 50 U/ml penicillin-G and 50 μg/ml streptomycin, supplemented with 7% fetal calf serum (FCS; BioWhittaker, Walkersville, MD) or 0·5–1% syngeneic normal mouse serum (NMS).

DC preparation

Splenic DC were prepared with modifications as described eslewhere.⁴ Briefly, spleens were dissociated mechanically in Hanks' balanced salt Solution (HBSS) with 5 mm EDTA, resuspended in complete medium, and cultured in plastic Petri dishes for 2 hr in medium with 1% NMS or 7% FCS. Non-adherent cells were removed and adherent cells were cultured overnight in medium with 0·5% NMS or 7% FCS, supplemented with 1 ng/ml recombinant mouse granulocyte–macrophage colony-stimulating factor, plus 1 μm SIINFEKL (pOVA_I) peptide (Invitrogen, Carlsbad, CA) defining the MHC I K^b-restricted ovalbumin (OVA) epitope OVA_257–264.²³ Endotoxin-free keyhole limpet haemocyanin (KLH; 100 μg/ml; Calbiochem, La Jolla, CA) was added to cultures set up in NMS. The next day, non-adherent cells were harvested, washed in HBSS, and analysed by flow cytometry using monoclonal antibodies purchased from BD Pharmingen, San Diego, CA, or eBioscience, San Diego, CA. Purity of DC (routinely 40–50%) was determined by staining with anti-CD11c (clone N418) and anti-MHC II (clone M5/114). Anti-FcγRIII/II (clone 2.4G2, purified from culture supernatant) was used to block non-specific antibody binding. Data were collected on a FACSCalibur flow cytometer and analysed with flowjo (Treestar, Ashland, OR) or cellquest (BD Biosciences, San Diego, CA) software.

Immunizations

Spleen cells were loaded with OVA protein (Sigma, St Louis, MO) as described previously.²⁴ Briefly, 100 × 10⁶ to 120 × 10⁶ splenocytes were prepared from B6 or Tap-deficient mice, erythrocytes were lysed, and cells were resuspended in 1 ml OVA in HBSS (10 mg/ml). After 10 min at 37° the cells were washed extensively in HBSS, irradiated (1000 rad) and resuspended at 15 × 10⁶ to 25 × 10⁶ cells in 200 μl HBSS for intravenous (i.v.) injection. Male or antigen-pulsed splenic DC (approximately 1 × 10⁵ CD11c⁺ cells) in HBSS were injected either intraperitoneally (i.p.; 200 μl) or subcutaneously (s.c.; 50 μl). Where indicated, mice were depleted of CD4⁺ cells by i.p. injection of 50 μg anti-CD4 antibody GK1.5 (purified in the laboratory from culture supernatant) 5 and 3 days before immunization. In some experiments, mice were injected s.c. with 300 ng mouse TNF-α (eBioscience) 7 hr before DC injection at the same site.²⁵ In other experiments, DC were injected in 50 μl HBSS with 10⁵ units of Universal Type I IFN (PBL Biomedical Laboratories, Piscataway, NJ), and a second s.c. injection (10⁵ units) of type I IFN was given at the same site 2 days later as described elsewhere.²

ELISPOT assay

ELISPOT assays specific for interleukin-2 (IL-2) and IFN-γ were performed as previously described.⁴ Briefly, twofold dilutions of immune lymph node or spleen cells were made in medium containing naive B6 splenocytes, so that the cell concentration remained constant at 5 × 10⁶ cells/ml. A total of 5 × 10⁵ cells per well were cultured in complete medium supplemented with 1% NMS, with or without antigen (1 μm pOVA_I, 100 μg/ml KLH, or 7% FCS). Plates were incubated at 37° for 18–20 hr. Spots were developed using biotinylated anti-IFN-γ (pOVA_I) or anti-IL-2 (KLH and FCS) followed by streptavidin-alkaline phosphatase and the substrate Vector Blue (Vector Laboratories, Burlingame, CA), and were counted using a CTL Analyzer with immunospot® software (CTL, Cleveland, OH). Data are shown as the mean of triplicate wells ± standard error. Background (minus antigen) values were consistently low, and were not subtracted.

In vivo elimination assay

Mice were depleted of NK cells by i.p. injection of 25 μl anti-asialo-GM1 (ASGM1) antiserum (Cedarlane, Burlington, NC), or were injected with 100 μg rabbit IgG as a control. Tap^−/− and B6 splenocytes were prepared in HBSS, erythrocytes were lysed, and the cells were stained differentially with carboxyfluorescein succinimidyl ester (CFSE), using 0·25 μm for Tap^−/− cells and 2·5 μm for B6 cells. After 10 min at 37° the cells were washed and mixed in similar proportions before injecting a total of 10 × 10⁶ to 15 × 10⁶ cells i.v. The ratio (r) of the two populations injected was calculated as [% CFSE^low (Tap^−/−) cells/% CFSE^high (B6) cells]. Spleens were harvested 20 hr later and analysed by flow cytometry for CFSE expression. Elimination of Tap^−/− cells was determined for each mouse by the following equation, gating on CFSE⁺ cells: % elimination = 100 × [(% B6 cells × r) − % Tap^−/− cells]/(% B6 cells × r).

Statistics

Responses of the relevant groups were compared using a one-tailed non-parametric Mann–Whitney U-test, as described in the figure legends.

Results

Injection of OVA-pulsed Tap-deficient cells activates NK cells, but the primary CD8⁺ T-cell response remains help-dependent

We first asked whether immunization with antigen-pulsed, MHC I-deficient cells could cross-prime CD8⁺ T-cell responses that were CD4⁺ T-cell-independent because of bystander help from activated NK cells.^17–19 B6 or Tap^−/− spleen cells were loaded with antigen by brief incubation with OVA protein, then injected i.v. into control or CD4-depleted B6 mice to generate CD8⁺ T-cell responses cross-primed by host antigen-presenting cells (APC) as described.²⁴ Spleens were collected 7 days after immunization and CD8⁺ T-cell responses specific for pOVA_I were measured by IFN-γ ELISPOT. As expected,^2,3,26 injection of OVA-pulsed B6 splenocytes generated a robust CD8⁺ T-cell response that was severely diminished in GK1.5-treated mice, confirming the help-dependence of this response (Fig. 1a). Injection of OVA-pulsed Tap^−/− cells into PBS-treated B6 mice cross-primed a response that was modest compared with the response to OVA-pulsed B6 cells. However, this response was significantly reduced in GK1.5-treated mice, showing that the CD8⁺ T-cell response to OVA-pulsed Tap^−/− cells was also help-dependent, despite any NK activation by the injected cells.

Figure 1.

CD8⁺ T-cell responses cross-primed by MHC I-deficient splenocytes are help-dependent. Control and CD4-depleted B6 mice were injected (a) with 15 × 10⁶ to 25 × 10⁶ ovalbumin (OVA) -pulsed splenocytes from B6 or Tap^−/− donors, or (b) with 1.2 × 10⁵ B6 or Tap^−/− male dendritic cells (DC). Splenic interferon-γ (IFN-γ) responses to ovalbumin_257–264 SIINFEKL peptide (pOVA_I) (a) or to the male antigen HY (b) were measured by ELISPOT 7 day later. Responses in the absence of peptide (a) or with female antigen-presenting cells (b) are shown in the lower graph. Each column represents the antigen-specific response of a single mouse; results are representative of two independent experiments. *P = 0·05, **P < 0·03, ***P < 0·02.

Similar results were obtained using lipopolysaccharide (LPS)-activated B6 or Tap^−/− male DC to prime/cross-prime CD8⁺ T-cell responses to the endogenously expressed HY male antigen. Previous studies showed that MHC I-deficient male cells cross-primed HY-specific CD8⁺ T-cell responses, though this response was small (approximately 10%) compared with the response primed by MHC I-sufficient cells.²⁷ We too found that HY-specific responses cross-primed by Tap^−/− DC were significantly weaker than responses primed by MHC I-sufficient DC, and were severely reduced or absent in CD4-depleted mice (Fig. 1b). Hence, in two different experimental models, CD8⁺ T-cell responses cross-primed by immunization with MHC I-deficient cells were strongly dependent on help from CD4⁺ T cells.

We confirmed that injection of Tap^−/− cells did in fact activate NK cells in our experiments, using a modified in vivo cytotoxicity assay²⁸ to ask whether Tap^−/− cells were eliminated by NK cells. B6 and Tap^−/− spleen cells were differentially labelled with CFSE to identify each population, and similar numbers of labelled cells were mixed and injected i.v. into B6 recipients previously injected with anti-ASGM1 antibody (to deplete NK cells) or control rabbit IgG. An aliquot was analysed by flow cytometry to determine the ratio of Tap^−/− and B6 cells injected. Twenty hours later, splenocytes from the recipient mice were analysed to measure the numbers of injected B6 and Tap^−/− cells, and so calculate the percentage of Tap^−/− cells eliminated (Fig. 2a). The ratio of Tap^−/− to B6 cells was substantially reduced in mice injected with control antibody, with up to 60% elimination of MHC I-deficient cells. In contrast, the ratio was relatively unchanged in mice injected with anti-ASGM1, indicating that NK cells were responsible for elimination of Tap^−/− cells. The T cell-depleted spleen cells from Tap^−/− mice were eliminated as effectively as whole spleen cell preparations (Fig. 2b), showing that elimination of Tap^−/− cells in vivo did not depend on donor Tap^−/− T-cell graft-versus-host responses to H-2^b class I MHC molecules. Flow cytometry experiments confirmed that NK cells were depleted by anti-ASGM1 treatment; the percentage of NK1.1⁺ and DX5⁺ splenocytes from mice injected with anti-ASGM1 48 hr previously was substantially reduced compared with cells from control mice (data not shown). Primary CD8⁺ T-cell responses cross-primed by Tap^−/− cells remained help-dependent even though the injected cells activated host NK cells. Our results are consistent with a recent study²⁰ showing that although NK cell activation by MHC I-deficient or allogeneic cells can augment CD8⁺ T-cell responses, it cannot support such responses in the absence of CD4⁺ T-cell help.

Figure 2.

Injection of MHC I-deficient splenocytes activates natural killer (NK) cell-mediated cytotoxicity. B6 mice were injected with anti-ASGM1 rabbit antiserum, or control rabbit IgG. The next day, they were injected with a 1 : 1 mixture of B6 and Tap^−/− cells from whole (a) or T-cell-depleted spleens (b), differentially labelled with carboxyfluorescein succinimidyl ester (CFSE). After 20 hr, the ratio of CFSE-labelled B6 versus Tap^−/− cells in the spleen was determined by flow cytometry.

DC stimulated in vitro by the inflammatory cytokines TNF-α and IFN-α do not prime help-independent CD8⁺ T-cell responses

We next tested whether inflammatory cytokines could restore primary CD8⁺ T-cell responses in the absence of CD4⁺ T-cell help, either through DC activation in vitro, or systemically after in vivo injection. We were particularly interested in cytokines linked to autoimmune disease, because dependence on help has been proposed as a mechanism for limiting CD8⁺ T-cell responses to autoantigens.²⁹ The cytokines TNF-α and IFN-α were chosen because they represent opposite ends of a spectrum of inflammatory cytokines³⁰ associated with autoimmune diseases such as rheumatoid arthritis (TNF-α) and systemic lupus erythematosus (IFN-α).

We have previously shown that immunization with small numbers of peptide-pulsed DC generates primary and secondary CD8⁺ T-cell responses that are strongly dependent on help from CD4⁺ T cells.^4,31,32 In the absence of help, antigen-specific responses measured by tetramer staining and IFN-γ secretion are severely reduced.^4,33 Experiments from several groups including our own showed that responses remained help-dependent despite maturation-associated up-regulation of adhesion and co-stimulatory molecules such intercellular adhesion molecule type 1, CD40 and CD86.^31,34,35 However, TNF-α^36,37 and IFN-α^36,38 both further stimulate DC maturation and activation, possibly tipping the balance to favour CD8⁺ T-cell responses. We therefore asked whether DC cultured overnight with TNF-α or IFN-α could stimulate help-independent responses.

Splenic DC were cultured in medium supplemented with NMS, plus KLH (to stimulate CD4⁺ T-cell help) and pOVA_I (to stimulate CD8⁺ T cells), with or without 10 ng/ml recombinant TNF-α as described.^36,37 The following day, DC were harvested, thoroughly washed to remove unbound antigen, then injected into mice previously injected with anti-CD4 antibody or PBS. Preliminary experiments confirmed that TNF-α-treated DC had elevated levels of CD40 and CD86 and increased secretion of IL-6 relative to DC cultured overnight in medium alone (data not shown). CD8⁺ T-cell responses to pOVA_I were measured ex vivo 7 days after immunization by IFN-γ ELISPOT (Fig. 3a). Antigen-pulsed DC primed substantial CD8⁺ T-cell responses to pOVA_I, but the response was diminished or absent in mice lacking CD4⁺ T cells. Immunization with TNF-α-treated DC generated pOVA_I-specific responses that were also highly help-dependent. Hence, although TNF-α increased DC expression of co-stimulatory molecules and cytokine secretion, it did not license them to prime help-independent CD8⁺ T-cell responses.

Figure 3.

Dendritic cells (DC) stimulated in vitro with tumour necrosis factor-α (TNF-α), interferon-α (IFN-α) or lipopolysaccharide (LPS), are unable to prime CD8⁺ T-cell responses in the absence of CD4⁺ T-cell help. Control or CD4-depleted B6 mice were injected with B6 DC cultured overnight (a) with keyhole limpet haemocyanin (KLH) plus ovalbumin_257–264 SIINFEKL peptide (pOVA_I), with or without 10 ng/ml TNF-α; (b) with fetal calf serum (FCS) plus pOVA_I, with or without 25 ng/ml type I IFN; or (c) with FCS plus pOVA_I, with or without 0.1 μg/ml LPS. The IFN-γ response to pOVA_I was measured by ELISPOT 7–8 days later. Results are representative of at least two independent experiments. *P = 0·05, **P = 0·03, ***P < 0·02.

In a second set of experiments, DC were cultured overnight with pOVA_I, plus FCS to stimulate CD4⁺ T-cell help,⁴ with or without IFN-α (25 ng/ml), then injected into untreated or CD4-depleted mice. When IFN-γ responses specific for pOVA_I were measured in the draining lymph nodes 5 days later, they were weak or absent in CD4-depleted mice compared with excellent responses in untreated mice (Fig. 3b). DC stimulation with IFN-α did not increase the response in untreated mice, or restore the response in CD4-depleted mice.

Finally, we tested whether DC activation with LPS could overcome the need for help. The DC were cultured overnight with FCS plus pOVA_I, with or without LPS (0·1 μg/ml) before injection into CD4-depleted and control mice. Both untreated and LPS-stimulated DC primed excellent IFN-γ responses to pOVA_I, (Fig. 3c), but again there was little if any response in CD4-depleted mice. We also confirmed that GK1.5 treatment abolished the CD4⁺ T-cell response (data not shown), measured by FCS-specific IL-2 spots.⁴ In our experimental model, primary CD8⁺ T-cell responses remained help-dependent even when the immunizing DC were activated by LPS.

Tissue conditioning by TNF-α fails to restore primary CD8⁺ T-cell responses in CD4-depleted mice

CD4⁺ T-cell-derived chemokines enhance recruitment of antigen-specific CD8⁺ T cells to lymph nodes, improving their chances of interaction with antigen-expressing DC.³⁹ Local tissue conditioning by TNF-α also increases lymphocyte recruitment to draining lymph nodes, and promotes CD4⁺ T-cell activation.²⁵ We therefore asked whether local tissue conditioning by TNF-α could reduce the need for CD4⁺ T-cell help in our experimental model. We first confirmed that s.c. injection of TNF-α increased cellularity in the draining lymph node at 24 hr in both control and CD4-depleted B6 mice (Fig. 4a). Lymph node cellularity was unaffected by GK1.5 treatment in both control and TNF-α-injected groups, as the depletion of CD4⁺ T cells in GK1.5-injected mice was balanced by an increase in B220⁺ cells (data not shown). We next injected control and CD4-depleted mice in the shoulder with PBS or TNF-α, followed 7 hr later by injection at the same site with DC cultured overnight in medium with FCS plus pOVA_I. Both PBS-injected and TNF-α-injected mice made IFN-γ responses to pOVA_I that were significantly diminished in CD4-depleted mice, indicating that localized TNF-α signalling was insufficient to drive CD8⁺ T-cell responses in the absence of help (Fig. 4b). Additionally, we observed no difference in the responses of control versus TNF-α-injected mice, measured either as the frequency of pOVA_I-specific cells (Fig. 4b), or as the total number of pOVA-specific CD8⁺ T cells in the draining lymph nodes (not shown). Although injection of TNF-α did temporarily increase lymph node cellularity, it did not restore primary CD8⁺ T-cell responses in CD4-depleted mice to the levels seen in CD4-sufficient mice.

Figure 4.

Injection of tumour necrosis factor-α (TNF-α) increases cellularity in the draining lymph nodes, but does not restore CD8⁺ T-cell responses in the absence of help. (a) Control or CD4-depleted B6 mice were injected subcutaneously (s.c.) with PBS or 300 ng TNF-α. Brachial and axillary draining lymph nodes were removed 24 hr later, pooled, counted and analysed by flow cytometry. Filled symbols indicate results from one experiment; open symbols represent two additional experiments. (b) Control or CD4-depleted B6 mice were injected s.c. with PBS or 300 ng TNF-α, and were injected 7 hr later at the same site with B6 denderitic cells pulsed with FCS plus ovalbumin_257–264 SIINFEKL peptide (pOVA_I). Five days later, draining lymph node cells were analysed for interferon-γ responses to pOVA_I. Data in (b) are representative of three independent experiments. *P < 0·02, **P < 0·003, ***P < 0·001.

CD4-depleted mice make primary CD8⁺ T-cell responses to peptide-pulsed DC when injected with IFN-α

Adoptive transfer experiments with IFNRI-deficient TCR transgenic CD8⁺ T cells have demonstrated that IFN-α can act directly on CD8⁺ T cells to promote expansion and survival in both virus infection and cross-priming experiments.^40–45 We therefore tested whether exogenous IFN-α could support help-independent CD8⁺ T-cell responses in our experimental model. Control and CD4-depleted B6 mice were injected s.c. with DC pulsed with FCS and pOVA_I as described above. One group of CD4-depleted mice was injected with antigen-pulsed DC plus 10⁵ IU recombinant IFN-α; these mice received a second injection of IFN-α 48 hr later as described.² Interferon-γ responses to pOVA_I, and IL-2 responses to FCS antigens, were measured in the draining lymph node 5 day later (Fig. 5). Control mice made substantial IFN-γ responses to pOVA_I that were severely diminished in CD4-depleted mice, confirming the help-dependence of the response. In contrast, CD4-depleted mice injected with IFN-α made pOVA_I-specific responses that were similar to those of control untreated mice. The weak response to FCS antigens in the IFN-α treated group confirmed the absence of help in these mice. Hence, injection of type I IFN supported the generation of help-independent primary CD8⁺ T-cell responses to cell-associated antigens.

Figure 5.

Injection of interferon-α (IFN-α) supports primary CD8⁺ T cell-responses in CD4-depleted mice. Control and CD4-depleted B6 mice were injected subcutaneously (s.c.) with B6 dendritic cells (DC) pulsed with FCS plus ovalbumin_257–264 SIINFEKL peptide (pOVA_I). One group of CD4-depleted mice (grey bars) was also injected at the same site with 10⁵ IU type I IFN on day 0 and day 2. Draining lymph node cells were assayed on day 5 for IFN-γ responses to pOVA_I and interleukin-2 (IL-2) responses to FCS. *P < 0·02.

Type I IFN is not required for primary CD8⁺ T-cell responses to peptide-pulsed DC

We speculated that IFN-α might be a critical downstream mediator of CD4⁺ T-cell help, in which case IFNR1-deficient mice should be unable to make primary CD8⁺ T-cell responses to peptide-pulsed DC. Alternatively, IFN-α might not be important for help, but might drive CD8⁺ T-cell expansion and/or survival by a different mechanism. We therefore investigated whether stimulation through the IFNR1 was required for responses primed by peptide-pulsed DC. Wild-type or IFNR1-deficient DC pulsed with pOVA_I plus FCS were injected into untreated or CD4-depleted mice. Both DC populations primed IFN-γ responses to pOVA_I, and IL-2 responses to FCS, that were almost completely abrogated in CD4-depleted mice (Fig. 6); hence, the help-dependent CD8⁺ T-cell responses did not require DC stimulation by type I IFN. To test whether responses primed by peptide-pulsed DC were at all dependent on stimulation through the IFNR1, we immunized IFNR1-deficient mice with peptide-pulsed IFNR1-deficient DC. Three of four IFNR1-deficient mice made excellent IFN-γ responses to pOVA_I, showing that CD8⁺ T-cell responses could be primed by peptide-pulsed DC in the complete absence of IFNR1 stimulation (data not shown). This suggests that type I IFN does not replace CD4⁺ T-cell help, but instead supports primary CD8⁺ T-cell responses through an alternative pathway.

Figure 6.

CD8⁺ T-cell responses primed by peptide-pulsed dendritic cells (DC) are independent of DC stimulation by type I interferon (IFN). Control or CD4-depleted B6 mice were immunized with B6 or type I IFN receptor 1 deficient (IFNRI^−/−) DC pulsed with FCS plus ovalbumin_257–264 SIINFEKL peptide (pOVA_I). Splenocytes were assayed on day 7 for IFN-γ responses to pOVA_I and interleukin-2 (IL-2) responses to FCS. *P < 0·02.

Discussion

In this study, we asked why primary CD8⁺ T-cell responses to non-pathogen, cell-associated antigens show such wide variation in their requirement for CD4⁺ T-cell help. We previously suggested that inflammation stimulated by injection of large numbers of apoptotic cells, and/or bystander help from NK cells activated by MHC I-deficient cells, might support CD8⁺ T-cell responses in the absence of CD4⁺ T cells,⁴ and we show here that the inflammatory cytokine IFN-α can indeed support primary CD8⁺ T-cell responses to peptide-pulsed DC responses in the absence of help. This is a specific effect of IFN-α, rather than a more general consequence of inflammation, because TNF-α does not support help-dependent responses, despite increasing lymph node cellularity. Type I IFN has now been shown to be a key mediator of T-cell responses induced by apoptotic cells.⁴⁶ and is likely to have been a critical component of the immunization strategy that originally demonstrated help-independent primary CD8⁺ T-cell responses to cell-associated antigens.⁷ We suggest that the help-dependence of primary CD8⁺ T-cell responses to such antigens may correlate inversely with the levels of endogenous type I IFN stimulated by different immunization protocols.

We found that NK cell activation by MHC I-deficient cells did not support help-independent primary CD8⁺ T-cell responses. This is consistent with a recent study showing that CD8⁺ T-cell responses cross-primed by OVA-transgenic, MHC I-deficient or allogeneic spleen cells remained help-dependent, even though NK cells were activated, and strongly enhanced both T-cell and B-cell responses in CD4-sufficient mice.²⁰ We speculate that the poor responses observed in CD4-sufficient mice to MHC I-deficient OVA-pulsed spleen cells or male DC could be the result of the limited amounts of antigen available for cross-presentation, compared with antigen levels in MHC I-deficient cells expressing OVA as a transgene. Natural killer cell-mediated direct or indirect inflammatory signals can drive help-independent CD8⁺ T-cell responses in some experimental models of cross-presentation; however, one early report of NK-mediated help for CD8⁺ T-cell responses primed by MHC I-deficient tumour cells¹⁹ was later shown to depend on NK cell stimulation by NKG2D molecules expressed at high levels on the tumours in question,¹⁸ in contrast with splenocytes or DC, which do not express high levels of NKG2D.

Type I IFN can affect CD8⁺ T-cell responses through multiple mechanisms, stimulating DC maturation and cross-presentation of exogenous antigens; increasing the number, effector functions, and anti-tumour activity of peptide-primed CD8⁺ T cells; driving CD8⁺ T-cell responses to pathogen infection; and stimulating IL-15 and IL-15Rα expression on myeloid cells, and so promoting CD8⁺ T-cell survival and renewal.^{2,36,40–45,47–50} Our results ruled out any significant effects on the immunizing DC; IFN-α-treated DC up-regulated co-stimulatory molecules, but primed responses that were indistinguishable from controls when help was available, and did not prime CD8⁺ T-cell responses in CD4-depleted mice. Although DC cross-presentation of exogenous antigens is strongly dependent on type I IFN,² the majority of our experiments involved immunization with peptide-pulsed DC, which prime CD8⁺ T-cell responses directly, without any need for further processing or cross-presentation.³² It therefore seems likely that IFN-α acted directly on the CD8⁺ T cells themselves, stimulating them to proliferate and survive, as previously demonstrated for CD8⁺ T-cell responses to pathogens and tumour antigens.^{40,41,43–45}

We suggest that the help-dependence of primary CD8⁺ T-cell responses to cell-associated antigens is linked to the levels of type I IFN stimulated by different immunization protocols. Immunization with relatively undamaged, peptide-pulsed DC stimulates minimal inflammation, priming strongly help-dependent responses⁴ that do not require IFNR1 signalling. Cross-primed CD8⁺ T-cell responses required type I IFN to trigger antigen processing by host APC^2,46 even if help is available; the observation that cross-primed CD8⁺ T-cell responses to OVA-loaded cells can be help-dependent^2,3 suggests that levels of type I IFN sufficient to stimulate antigen cross-presentation by host APC are not necessarily high enough to drive CD8⁺ T-cell proliferation in the absence of help. Finally, injection of antigen-expressing cells severely damaged by irradiation or freeze–thawing may stimulate high levels of type I IFN that drive both cross-presentation and CD8⁺ T-cell expansion and survival in the absence of help.^7,46 If the help-dependence of primary CD8⁺ T-cell responses correlates with levels of type I IFN, which in turn correlate with the extent of damage to the cells used for immunization, then differences in cell preparation protocols (even within one laboratory) may critically affect the extent to which primary CD8⁺ T-cell responses depend on help.

It is now widely recognized that inflammatory conditions during priming are critically important for the quality of the CD8⁺ T-cell response.^51,52 Both IFNR1-deficient and IL-12p40-deficient mice made primary CD8⁺ T-cell responses to peptide-pulsed, syngeneic DC (this study, and A.M.L. unpublished results), suggesting that the inflammatory cytokines IL-12 and type I IFN are not required in our experimental model. Because type I IFN and IL-12 can both provide critical ‘third signals’ that contribute to the development of CD8⁺ T-cell effector functions,⁵³ unhelped primary CD8⁺ T-cell responses supported by type I IFN may have more potent effector functions than help-dependent responses primed by DC in the absence of overt inflammation. They may also be vulnerable to TRAIL-mediated apoptosis upon restimulation,⁵⁴ although the generation of TRAIL-sensitive cells may depend on additional stimulation through receptors for endogenous danger signals and NK cell activation.^20,46 The quality of unhelped CD8⁺ T-cell responses supported by type I IFN will be the focus of future studies.

The help-dependence of CD8⁺ T-cell responses has been proposed as a key mechanism for maintaining self tolerance.²⁹ If so, elevated levels of inflammatory cytokines that support CD8⁺ T-cell responses in the absence of help might contribute to the initiation and pathology of autoimmune disease. Our results suggest that individuals with elevated levels of TNF-α (e.g. patients with rheumatoid arthritis) may not have any increased risk of CD8⁺ T-cell responses to self antigens. However, help-independent CD8⁺ T-cell responses to self antigens may contribute to disease initiation and progression in individuals with autoimmune diseases (e.g. systemic lupus erythematosus) that are associated with high levels of type I IFN,⁵⁵ and could conceivably be a concern for conditions that are treated with exogenous type I IFN.

Glossary

Abbreviations

APC

antigen-presenting cells

ASGM1

asialo-GM1

B6

C57BL/6

CFSE

carboxyfluorescein succinimidyl ester

DC

dendritic cells

FCS

fetal calf serum

HBSS

Hanks' balanced salt solution

IFNR1

type I IFN receptor 1

IFN-α

type I interferon-α

IL

interleukin

i.p.

intraperitoneally

i.v.

intravenously

KLH

keyhole limpet haemocyanin

LPS

lipopolysaccharide

NK

natural killer

NMS

normal mouse serum

OVA

ovalbumin

pOVA_I

OVA_257–264 SIINFEKL peptide

s.c.

subcutaneously

TNF-α

tumour necrosis factor-α

Disclosures

The authors have no conflicts of interest to disclose.